In order to reduce the size of inverted matrices in BGFA, I implemented some complicated indexing that selects sub-model matrices that are independent of the rest, and processes their likelihood separately. The code to do this is a nightmare both in transformed data{} and model{} but made a huge speed difference. Think about ways to simplify!
In order to reduce the size of inverted matrices in BGFA, I implemented some complicated indexing that selects sub-model matrices that are independent of the rest, and processes their likelihood separately. The code to do this is a nightmare both in transformed data{} and model{} but made a huge speed difference. Think about ways to simplify!